Advanced ROI Measurement Tactics, Part II: Keyword Assist
This is my second post in a series about SEM tactics that lead to greater ROI from search campaigns. See my first post in the series on online-offline revenue.
This implementation allows users to allocate conversions back to keywords prior to the keywords that led to a conversion. In other words, conversions are allocated back to SearchCenter campaigns using either a First, Last, or Single-Visit allocation method.
For example, imagine a company that sells rugs. When people are thinking of purchasing a rug, they likely begin searching for the word “rug” or “area rug.” Then, as they narrow down their options, they may search for “Persian area rug” or “modern area rug.” The second, more specific keyword is what usually gets credit for the sale in most systems, while the first keyword is all but ignored. An SEM marketer might be inclined to devalue the first keyword, “rug,” because it is expensive — yet it’s creating sales by being the searcher’s first contact with the company.
The CVP custom solution will also allow allocation of conversions across campaigns from multiple visits. Then, a report can be run to show conversions, side-by-side, for both Last and Full-Linear allocation.
That being said, studies we’ve done across our customers show that it takes about 1.2 keywords being clicked on to make a sale. So, while more than one keyword is being clicked, it doesn’t happen very often. In order to make a tactic like this worthwhile, you want to look at two types of keywords: very broad consideration keywords that are high volume and high cost, and brand keywords.
These tactics can be extremely worthwhile in terms of ROI, but they take more consideration than the standard implementation. Bear in mind that they generally require a consulting engagement to get them set up in the most efficient manner.
Next time, I’ll continue to look at integrations of measurement solutions by introducing how to use Vista rules for cost-of-goods-sold analysis and cross-channel marketing effects.